good bye 'speed density', hello maf...

I converted my tbi over to run entirely off of a maf with the map sensor handling the other things like ae and the timing tables. Its one of the huge ls1 mafs off of a late model yukon. Got it for $25 or so +s/h. Reads up to something like 450 grams/sec and is over 3.5 in. in diameter. I have to say that its running really well considering ive only done two chips. In those two chips it runs better then the tbi did with 30. Im really starting to fall in love with it.

As far as the mods go, this ones a solid 10+ on the PITA scale. Involved lots of math, engine theory, LOTS of programming in assembly, and some basic electronics design. The code itself was inspired by the code running on the '165 $6E maf ecms, but once it was all done, it looks more like the fueling side of the $32 code.

anyway, heres the pic of the maf setup.

 

heres the converter box I have to use to interface it with my ecm

 

thats cool but id be afraid to screw mine up, im not that good with that stuff

 

Heh, I was, too. I had to do everything from scratch so I was expecting it to not even work at all. I had to do the work that would normally be done by a team of automotive engineers with a well equiped test facility. All I had to design my setup was my ecm test bench and my TI-89 calculator. *** bless texas instruments...

 

cool! i plan on getting a second third gen to toy with. my camaro is in great shape (body wise anyway), my engine has some small repairs needed.

 

Awesome Dimented! I've been watching your thread on the Prom board about it, really glad to see it worked out well. You think it's got more on top, better driveability? Where does it outperform the speed density setup best? This is a really cool project. Gotta love those Ti89's

 

DIY Snob!

Just kidding.

What kind of performance gain are you seeing so far? What do you expect?

What is that pulley on the far left of the first picture?

 

AC delete pulley

It's looking good dimented. Don't think you'll see any other TBI motors looking quite like yours anymore!

 

That pully is an A/C delete pully and behind it you can see a non-A/C heater box.

As far as the MAF goes, one of the real reasons I decided to go with it is that the maf takes alot of the variables out of the equation.

This has been a hunch of mine for a while:

With speed density, the basic theory is as follows: you can estimate the mass of air in a cylinder of a motor by having the absolute pressure, volume of the cylinder, and the ambient temperature. Basically with the ideal gas laws and the density of air you get the mass of air. In some basic form, you can calculate the volume of air relative to atmospheric pressure and then with the density of air and the volumetric efficency of the motor you get an esitimate of the mass oif air. That mass of air along with the air/fuel ratio will give you the needed mass of fuel. Divide that by the total flowrate of fuel provided by the injector and you get a time that the injector should be on for (I think all of that is somewhat correct). Seems simple enough...

With tbi, you dont have just air. What you have is air and fuel vapor with a small volume of liquid fuel since the gas is squirted in with the air right at the tbi rather then at the cylinder. What the vaporized fuel does is displace the air since its a vapor as well. This means that if you just use the idealized laws, your calculations for the mass of the air in each cylinder will be off since a good deal of the fuel mixed in with the air since its a vapor. This means that there will be some sort of corrections needed to account for this. If you look at the tbi code, it looks wierd. The inverse afr really isnt an inverse afr like youd expect. It looks like it, but its sort of fudged. The calculations to estimate the airflow through the motor use silly map terms and rpm scalars rather then just using ve table, rpm, and the map readings. The bpw calc also looks kinda funky. I always thought that it was just the engineers doing weird stuff, but when I tought of the possible issues with fuel dynamics, it sort of made sense that they would need to fudge things a bit.

To me all of this means that the tbi system will be highly succeptable to the type of intake used and the engines temperature. This makes tuning more difficult since most of it just becomes iterative testing to find what works. At least with an mpfi system there is some theory to hang the fueling algorithm on. With tbi it all seems like something resembling SD with lots of fudge factor tossed in.

This is all just a bunch of theory, though, so I could be completly wrong.

The fueling with maf can be boiled down to: injector duty cycle = (grams/sec of airflow)/(grams/sec of total fuel flow) x (1/air fuel ratio). Multiply the result of the total time the injector has to fire, and you have the base pulsewith for the injector. Much simpler, which is why engineers (*cough* *cough* me *cough*) like and use it so much.

With maf, what I hope is that it takes alot of the variables like intake charge temp, engine temp, air temp, altitude, ve, timing, the tide, the phases of the moon, etc. and clears that right off the table. In theory (crosses fingers) the maf should be internally temperature compensated since it has its own internal temp sendor and it should also be immune to engine mods and timing changes since all it does is measure airflow. With SD, you change the timing, you need to retune it. Change the cam and heads, and you really need to tune it. This isnt to say that maf is the be-all, end all, though.

With maf your not tuning, all your doing is calibrating. Theres no insight at all. Just plugging numbers in a table to get the desired results. With SD, there are lots of subtle cues as to whats going on in the motor so you get much better insite into whats happening in the motor as you make changes. its also more intuative and has better resolution. With these older ecms, the maf is somewhat coarse since the resolution isnt that good. With pulse accumulation, alot of this has been overcome, but thats only available with the new pcms in use with the lt1 and ls1 platforms.

To get back to the origional question of improvements, the car seems to run more consistantly. I used an older VE table to set up my tune after my car sat for several months with a dead transmission but I radically changed my timing tables so it didnt run well at all. The maf seems to provide good performance at all temperature ranges, not just with the motor warm. Im sure with a WB I could sort all this out, but that takes time, not to mention that I have to wait for the seasons to change to see how the car runs at various temperatures.

Anyway, Im hoping itll work out. One major issue is varnishing. With all that gasoline floating around, the maf could collect alot of varnish and become a maintainance item. If the maf doesnt work out, at least Ill still have a cool CAI.

 

No wiring diagrams or anything?

 

I was pretty much flying by the seat of my pants on this one when I wired it up, so no schematics or diagrams. Pretty much, the maf needs +12V (I took this off after the transient voltage suppressor and diode to help protect the MAF) and more importantly, a dedicated ground that ties the maf, frequency converter, and ecm together onto a common ground. if there are voltage differences between the grounds, itll effect the voltage output and could effect the triggering of the LM2917. The maf frequency output works like an ALDL cable. Theres a transister that grounds a drop down resister hooked to constant +5 volts to generate the waveform.

I copied the schematics for the converter out of a datasheet for the LM2917 (www.jameco.com) and the power conditioning circuit from RBob. The schematics I used where for a voltage driven tach used with a dist. with breaker points. It has a 10k resister and a diode to generate a .5 volt reference for the trigger. This provides noise immunity since it will now trigger around .5 volts rather then ground.

I wish I could say more, but I gave up on schematics because I had to change things around so much to get to the final design.

 

It might seem like im hiding stuff but the truth is I cant even remember all the stuff I did. I didnt quite realize the full scope of what had to be done to essentially redo a sizable portion of the stock fueling algorithm and interface the ecm with a sensor that was made for a late model 32 bit pcm.

When I started I was just toying with the idea, but not too serious. Then I was like: 'well, lets do some basic stuff first just so I can have the maf in my datalogs'. After that I was like: 'well, why not just scrap the stock fueling code and copy/past this this $6E code in. Then it was: 'well, this $6E code kinda sucks, I want to make my own code.'

After that it was:

'Man, Im going to need error handling, a 24 bit filtering routine for the 16 bit airfow term, 32 bit mult. routine for the BPW calcs., 16 bit lookup routine with interpolation to handle a 16 bit inverse AFR table and a full 65 line 16 bit maf table, lookup routine, bpw calc. routine, not to mention that Ill have to interface it all with the stock code. Then I have to test it all and do the fine tuning/tweaking/debugging. Oh, wait, I still havnt figured out how to wire it or get the signal in...'

It was one of those projects that just got bigger and bigger and bigger and I couldnt put it down. I sure hope it works out over the long run.

 

I don't mean to hijack, but are you running a stock TBI? The readers rides is still down and I'm curious about your TB hat. I've done TPI CAI but I can't find a hat to finish the install

 

Bravo!!!! This kind of stuff is the coolest. It's the power of source coding in action. With code, the sky is the limit, well your brain is. Need a shift light, no problem, need traction control, no problem, need to control your license plate changer and passenger seat ejector, no problem.

As soon as I get out of school (May) I'm going to dive in. I have a good electronics teacher that knows all about source code and he's going to help me out.

 

Lots of learning here too.

Thanks D!

 

Kewl Stuff.

Congrats on getting it working ! !!

 

Great work Dimented!

 

A lot of your theory is a good collective of past discussion of theory so I think you're theory is essentially right on.

Thats sounds hilarious as I re-read it, but fug it, I'll leave it for others to laugh at as well.

Cool deal man. Glad it works. Let us know how the vapor issue effects the sensor.

 

www.superiorairflow.com Its the composit one. They put baffles in them so it can make mounting a pain.

EDIT: they dont make them anymore... but, they still have the expensive aluminum ones. Do you have an mpfi system or tbi? Theres a baffle in it that has to be cut out in order to get it to clear the injector pod on a tbi and with aluminum it would be a real PITA to get out.

 

Thanks.

I the maf mostly cailibrated and the blms are around 125-130. The car definatly runs good enough to drive. The idles still a little quirky. At ilde it only pulls around 5 g/s of airflow so the resolution problems and low velocity probably makes the fueling fluctuate.

Its wierd to not have the fueling be coupled to timing and engine parameters like it is with speed density. Before when I set the timing Id have to go back and redo the VE tables, now its jsut set teh timing table as I please and go.

 

I was just thinking...

If an ex-GM employee or current that worked on the original tuning of the tbi system saw all of this and visited these boards - They would probably be amazed at how much time and effort and progress people have put into modifying or improving the system they designed. But I bet they would be pretty impressed too.

ahhh, deep thoughts.. by snflupigus

 

That's really good stuff Dimented, I want to see how this works out in the long run. Your theory makes perfect sense that it should work better over speed density on a wet flow system like this. Keep us updated on your progress with it.

 

After a weekend, 100 miles of driving/tuning, $10 dollars of gas, and a bunch of chips, as far Im concerned, the MAF side of the tuning is done for now. Gotta get my truck up n' running so Im going to start concentrating on that.

The MAF I used is definatly way too big for my asthmatic 350. At very low flow rates it was bucking pretty good at cruise with the torque converter locked up. I had to add a variable filter coefficient for the MAF to filter the living snot out of it at idle and cruise while still having a fast response at higher rpms/loads. Helped eliminate most of the bucking, but it still does it a little. Also did worlds for the idle. Almost idles normal now. A bosch maf from a tpi system would have been more appropriatly sized but theres no way in hell im paying 150-200 dollars for that turd. Ill never use another bosch product after all the headaches Ive had from thier stuff.

Other then that it runs well. I probably could add more code to refine it, but my C3 is bursting at the seams already Not going to get much more stuff in there...

 

Speaking of theory, I was googling, and I saw what looked like an old SAE article online with the description that went something like this:

'While SD is recommended for MPFI systems, with TBI, MAF is more appropriate as with a speed density system it will only be an approximation, the reason for this is...'

but the friggin link was dead!!! I really would love to see the rest of that article. I have a feeling there might have been some usful info there that could help everyone running tbi with the stock SD system.

 

I would like to see your BIN & ECM files unless you are keeping them a secret. I am still trying to piece togather how you did it. I have thought about using the MAF sensor from a ford because I believe it uses a straight voltage signal.

 

PM me with your email addr and Ill send you the commented source code. As a disclamer, I think its error free but I cant garantee anything. Also, while the source code is functionally correct, all the stuff in the tables is just there to test the code. If your still interested in the LS1 maf, theres some stuff that can help. The MAF circuit is from one of the aussie service manuals and the circuit for the converter is from national semiconductors data sheet. Its commonly available and they give all the necessary info in there. The only real concerns are the ripple voltage and the temperature dependance of the timing cap. Want to choose one thats stable with varying temps or youll get some really screwy readings.

 

Heres the schematic for the converter. The stuff circled and crossed out is filtering that you dont want to use. Also, if you note in the pic of the maf schematics, the maf needs +5 volts, so youll need a +5 volt voltage regulator and a 10k resister.

 

Youll also want to use some power input protection circuitry. I cant find the schematics at the moment. Ill have to dig around a bit.

I really wish I kept better track of what I did...

 

you built it like the f'n russian space station.... LOL - somethin breaks and they gotta wake up 15 80 year old scientists and ask em to remember how they put it together... All because they were afraid someone would steal the technology... LOL

 

So, is your conclusion that this would make tuning a TBI vehicle much easier?

How would what you did compare with using the MAF setup on a P4?

I saw in a mag somewhere (CRS again ) where they had mounted the MAF directly above the TB. Of course it was MPFI, but..........

You mentioned the MAF being too large for your engine. I have the perfect guinea pig; more cubes, and P4

 

hehe... really I jsut took everyone elses ideas and schematics and made them work together in the way I wanted. Thats alot more like the russian way of doing things then Id like to admit. Im not really good at coming up with fresh whole new designs, Im just good at bumming the info off of everyone else. This site and the people here rock.


EDIT: I see what your saying, now. Too early when I replied.

 

Possibly, although part throttle and idle seem to be tricky to tune. Im sure a good cam could possibly spell a setup thats good at WOT but is a hair puller to tune at idle. Probably using some form of speed density for low loads or even a table of low load pulsewidths verses map and rpm could help, but im fresh out of room processing wise in my ecm.

The LS1 mafs are cheap at the moment so its not a high dollar project. Worth trying if you like to experiment. If you had a maf tpi using an ls1 maf would be plug n' play. Just need the converter.

 

Will using the ECMs +5 volt reference voltage not work? Are there worries of overloading the VR circuitry in the computer?

 

Yes, using the +5V reference would work. My converter is external to the ecm so I jsut put a small voltage regulator in with the rest of the circuitry. It minimized the number of wires I had to run. The MAF can and is normally fed with battery power but as a precaution, I routed the MAFs power after the blocking diode and the transient voltage suppressor that im using for teh converter to help provide extra protection.

 

:hail:

*edit - that's supposed to be the guy bowing down, not the red x of death.

 

Thanks for the link.

 

You mentioned the fact of the MAF being too large for your engine. Do you think it would be easier to tune with a larger cid, larger TB, and singleplane vs dualplane. Really, I know you would just be guessing at this point

More room available in a P4? Plus, would the faster baud rate of the P4 help?

Idle and part throttle (especially low load) are tricky to tune with any type of injection, no?

 

I think that with a larger engine the maf may be easier to tune. Not only is there more resolution as the flow increases, but the velocity of the airflow will be greater.

I finally got an aircleaner for it. One of those K&N cone airfilters. Got the largest one that would fit. The gauze doesnt flow that well with oil all over it so the filter blocks alot of the ambient airflow. The bucking at p/t is gone. One thing that I found very useful was the variable filter coef. That seemed to be able to compensate for alot of stuff. Without it and the maf open, the car was nearly undrivable. That could come in handy with a cam.

I think processing wise the P4 would be better. I can have the core of the fueling but there isnt enough time to allow for more code to take care of teh subtle details that make a good running system into a great system. I probably could have saved time by making the code more computer friendly like it was origionally, but the interface was really sloppy with all the scalars and seperate tables and stuff. All it is now is a large maf table and one AFR table. Much cleaner and easier to tune.

Its interesting at WOT. The car doesnt feel as fast now that its more smooth and linear. With the SD the power seemed to come in surges and waves. Hopefully this weekend I can get out and do some plug reading to see how the AFRs are at WOT. The O2 shows that its on the rich side but I dont know how trustworthy it is. RBob had a good point. With the MAF able to see the airflow most of teh AE should be able to be eliminated as the maf will add the fuel on its own. If the maf really can adiquatly do that, it would make the maf a truely powerful tool. Without the AE and the VE tables and all the associated temperature corrections, there arnt many variables left to tune. Just leaves the airflow calibration table. Kinda scary that all the fueling hinges on that.

 

This is some good work, I hope the results in the end are positive.

 

Awesome work, not just get it running well enough and you can use my car as a test mule for something with a little more power . I've also got some pretty extensive data-aq that might help speed things up.
I'm impressed you got it to even run the first time it was bolted on!

 

Thanks... Im surprised how well it does run. It really exceded my ecpectations. Even when cold it still runs decent. The maf does vary a couple of percent with temperature, but its nothing major. The things I still have to address are: timing (not related to the MAF), warm startup, and checking the high flow rates to make sure theyre accurate. On startup with the motor hot, it sort of bogs down when it catches. I think it may have too much choke. It also has some little quirks here and there but I dont know if Ill bother chasing the tiny things quite yet. With no VE tables, temperature corrections, or anything like that, theres hardly any variables left. In two weeks Ive gotten the car to run just about the way I want it. Tuning isnt intuative at all, though. Just trying stuff untill it works right.

On a side note, the transient operation is interesting. That part about not needing MAP AE doesnt quite hold true as I found out through testing. It needs the full ammount of MAP AE or slow filtered tps AE, and a touch of short TPS AE to work properly. Basically when I slammed the gas down the engine would stumble from being lean. The MAF seems to be able to compensate for manifold filling, but it still needs a very short burst of AE to not momentarily stumble on sudden tip in. This is probably needed to compensate for any delays. I automatically load the unfiltered maf when the throttle is opening, but that doesnt seem to even be fast enough to keep up.

This also seems to be a problem with the taurus that I was using. Slam the gas down and it hesitates. Normal person wouldnt notice it, let alone slam the gas down from a standing start. If I slam it down with the engine cold, it just about stalls out, but I suspect its for different reasons then my car.

 

The MAF has a transient flow delay based on airflow since air is a compressible. This is why MAF is great for "adapting" but it's not something you can rely on for fast AE. Using the TPS, RPM, and MAF delta flow you could come up with a scientific alogorithm to know exactly how much AE is required. This would all assume you're using a fixed volume intake .
It's the same reason why closed loop uses a PID alogorithm to adjust for the transfer times. It might seem like complex math but you can get away with a simple PID based on the above info that'll give you only 1 table and 2 variables to spit out an AE pulse that you ADD to the sync accum. Don't bother with async, RBob found it works best to just get rid of async and stick to sync. Takes some code work but man is it sweet. MUCH smoother than async fueling because of that whole dropping pulse widths when the timers overlap thing .
I like MAF because it does so much and with good code it's a tinker tuners dream since it gives you time to focus on other things like timing and lean mode.
How are you dealing with the timing? Do you still have a MAP sensor or is that Lv8 based?

 

That reminds me. About that compressibility thing. I remember seeing brief MAP readings of 104.4 kpa or whatever teh max reading it can have after the throttle opens suddenly. Really says something about how much inertia a column of air can have. For a short time, my motor actually runs under boost after I hit the gas.

I taked with RBob and I am going to convert over to using sync AE, but Ill postpone it untill I have a WB (finally getting close to buying one). Just too much of a pain tuning it cave man style.

I still use the MAP for timing, lookups for open loop AFR, etc. No sense in not using it for that. Thats what the MAP is good at. Really I just use the MAF for the fuel calcs. Most other things are done using the MAP. The LV8 seemed like too much of a pain.

 

Well, hard to believe its been over a year...

Im planning on using MAF again, although this time with a PCM. With the 8-bit A/D on the old C3, the resolution was really, really bad, especially at idle. My AFR choices at idle where:12.8, 13.7, 15.1. It would only hit those AFRs, and that was it due to the coarseness of the airflow readings from the A/D. It made for a really flakey idle. It was ok under modest loads, but in cruise with single digit airflows it would start bucking with the TCC engaged

With the P6, it actually has inputs that can accept and accuratly report (relatively) high frequency FM inputs up to around 40 kHz. With the new system, the MAF can just plug right into the PCM rather then having to have a converter and itll easily read in the 0-12 kHz FM signal without alot of processor overhead.

Should provide resolution down to around .05% or so of the reported airflow. It'll be interesting to see how a real MAF system runs.

 

This will also be a patch so it should run on any P6, provided you dont need EGR or the TOS input for the E-trans'.

 

Just out of curiosity, is there any interest for a patch like this for the TBI PCMs? Ill make it more general, rather then specific to my car, if there is. The patch also maintains the SD code intact and has an option to use either.